Paul M Clements

Many of the current crop of 3D rigs alter the Interocular (IO) distance by moving only one of the two cameras. This can be observed in Mirror Rigs using a Beam Splitter as well as simple side by side rigs. The main reason for doing this is to reduce manufacturing costs. Some might argue that fixing one of the cameras in position, such as on an overhanging mirror rig, means that the tolerances being used to keep that camera in position makes moving it, side to side, difficult. Not to mention more moving parts, means more to fail or to misalign. Whilst these points are valid there is the consideration that any moves using an offset IO could result in the left and right eye doing something different than what we are used to seeing in a 2D shot.

For example, if the left eye is central to a tripod and we pan, the left eye will turn less than the right eye, meaning instead of spinning around on a central point the stereoscopic effect pivots on the left eye and the right eye does a greater turn in an arc around the left. This results with an effect for the viewer of moving forward whilst turning their head (Try keeping your left eye in the same position and turning 90 degrees) . See the example beneath.

Non-centralized Interocular problem

So the question is does this matter? In truth, probably not. We don’t always pan a 2D camera with the film plain or sensor square in the centre of the tripod head (Front to back). And as an audience we never question whether that tripod pivoted precisely on a central point, it simply never enters the mind when watching a film or television. As such it is unlikely that we would do so with stereoscopic imagery, especially since the difference in both eyes is often very small.

Where a lack of centralization may be a problem is in the use of motion control rigs. Setting up a precise and detailed move using previz, only to discover that one eye is going to be off centre, could mean the move misses it’s intended marks. This would be even more relevant if the MoCo move requires motorized IO changes during the shot. Plotting both eyes in previz would be more work and could be slightly awkard as the two cameras paths would not match exactly. By contrast a centralized rig would allow the previz designer to simply refer to the centre of both cameras/eyes as he or she would a single camera and the field of view of both eyes changing IO as he or she would a zoom lens

For my mind a mirror rig would so rarely require such precision, due to the IO distances being so small, that adding the extra mechanisms into the design would only serve to make the rig even more unweildy than it already is as well as add more cost to the design and manufacturing. Although if it can be achieved without crippling the cost of production or making it’s operation less reliable, or more difficult, then as an addition to the rig’s design this would be a bonus over the current crop. With side by side rigs, however, where the IO distance can be significantly larger, I would argue it is a must.

When designing and developing a steroscopic camera rig using a beam splitter, the orientation of the mirrored camera to the camera that sees through the mirror is very important, especially when using digital cameras. The main issue concerns electronic shutters and the effects caused when the mirrored image is flipped or flopped (flipped vertically). Skew is one of the biggest problems with electronic shutters such as that in the Red One. Skew causes objects to look as though they have tilted to one side like the ‘Leaning tower of Piza’. It occurs when a camera is panned or more commonly when an object moves across the frame quickly. An example of which would be a bus or train moving in front of a static background. When shooting in 2D it is often easy to ignore or relatively simple to correct with the right tools. Get it wrong when shooting stereoscopically and you’ll be spending a very long time in post fixing the issue and quite probably coming to the conclusion that the shot is next to useless.

In the following diagrams are four different setups. Each setup has two images beneath, which illustrate what the cameras see, as well as a third image which shows the mirrored camera (Camera 2) after the image has been flipped or flopped. If you were to combine what Camera 1 sees and what Camera 2 (once flipped or flopped) sees this would give us our stereoscopic image. There are eight figures which demonstrate how each mirror rig setup would see an object (In this case a letter ‘R’). The first figure of each setup shows a static shot and the second shows how the object would look if the camera was panned left to right or the object in shot moved from camera right to camera left rapidly.

Read to the bottom to see the conclusion.

Beam splitter 3D Stereoscopic camera rigs and the importance of correct camera orientation highlighting the issue of skew using digital cameras

As we can see Setup 1 and Setup 4 when flipped have the object ‘R’ skewing in the same directions. As such these two methods represent the best solution. There is the added benefit that both camera’s sensors scan top to bottom in the same direction. We can see this because the top of the ‘R’ is at the top of each image before being flipped. Because of this, other unwanted effects that electronic shutters create will also match. An example would be a photographer standing in frame pointing a stills camera and using a flash. When the flash goes off an electronic shutter might see that flash over two frames. The consequence is that half the flash is visible on one frame overlapping the shot and half is visible in another frame. But since both cameras scan the same vertically they will both match (Perhaps not ideal but significantly better than Setup 3 and 4 would be).

So what’s wrong with these electronic shutters then, why do they have this weakness? Well essentially it’s due to the “Read/Write rate” of the sensor. Like a television or computer monitor the sensor starts in the top left and reads the first line of horizontal pixels across to the right. Once it reaches the far right edge it goes back to the left hand side and reads the next line of pixels beneath this, repeating over and over again until it reaches the very bottom of the sensor. This makes up a single frame of the footage. It then goes back up to the top left where it starts the process again to produce the next frame. A fast moving object, or the frame being panned, is moving faster than the Read/Write rate which means that each line as it is recorded from top to bottom is seeing the moving object or moving frame slightly further into shot than the above line of pixels. As such they do not line up vertically. The result is skew.

Will a faster Read/Write rate negate this issue? It will improve it considerably but it is likely that such artifacts will still be obvious to some degree in any electronic shutter.

Why not just use film then which doesn’t have this issue? Well it does, although not to the extent of most electronic shutters. But the real reason is because film has other problems which makes stereoscopy tricky. The size of the rig is one thing but tiny artifacts that appear on film as well as differing position of grain would mean that the left and right eye see slightly different shots. Having clean, noise free, digital footage is ideal and is in fact one of the reasons that stereoscopic cinematography is making a comeback since digital cinema cameras are improving rapidly.

Which is the better setup, 1 or 4? It’s really a personal choice as they both have their pro’s and con’s. With setup 1 in position on a tripod the rig looks very similar to an operator as a normal camera might. It doesn’t appear at first sight to be so cumbersome. But if the operator wants to tilt down or do long pans it will likely interfere with the tripod. Furthermore positioning the camera at floor level is difficult. Distributing the weight of the rig over a tripod is also problematic because it is always going to be front heavy. Putting weights at the rear of the rig to counter balance is common practice. The plus side of the design is that it uses less material to make. Setup 4 requires a lot of fabrication to hold the camera securely in such a position. However it can be put at floor level and does not interfere with the tripod in any way. When on a tripod the rig can be manouvered so that the balance of the head is perfect.

In conclusion, if you are hiring, buying or even producing a stereoscopic rig and intend to use a beam splitter either Setup 1 or 4 is the ideal choice.

Red classes their 2K windowed sensor (2048 x 1152 pixels) as being Super 16mm. It occured to me however to examine whether standard 16mm lenses would be useable with the RedOne given the difference in aspect ratios between the two formats. In a nutshell normal 16mm lenses may have no problem covering the 2K area of the sensor. And some may argue they are a better fit.

The following diagram illustrates the physical area of 16mm (10.26 by 7.49mm) as well as the physical area of S16mm (12.51 x 7.41mm). There are two circles of coverage of both formats. Finally the portion of Red’s sensor that is used when shooting 2K windowed can be seen to fit inside the 16mm’s circle of coverage.

If you click on the name of each item in the key you can hide them in the diagram so as to get a clearer understanding of what you are looking at.

The diagonal from corner to corner of 16mm is 12.70mm. So the minimum diameter of a 16mm lenses circle of coverage ought to be 12.70 or greater in order for the image to be useable. By comparison the diagonal of the 2K area of the red sensor is 12.65mm; 0.05mm less than the minimum 16mm circle of coverage’s diameter.

Finally, even if a standard 16mm lens should have issues covering the 2K area of the sensor they will still be more than capable of covering full HD, since the area of the sensor used for 1080p (1920 Pixels by 1080 Pixels) is easily smaller than standard 16mm (Footage would require batch cropping in RedCine from the 2K source although future options for Windowed 1080p may become available).

The result of the above means that standard 16mm lenses, such as the Cooke 9-50mm T2.5 or Zeiss 10-100mm T2.0, do not need to undergo modifications for use with S16mm which lessens the width of the lens as well as lowers the T-Stop and can on some lenses lower the general quality of the image. It also means that there are further lens options for shooters to choose from, and due to the popularity of S16mm or 35mm, these standard lenses may very well offer a better price perfomance ratio.

_______________________________________

Disclaimer: In the interest of being mathematically correct and fair I should illustrate how the size of the 2K sensor area was arrived at and whether any part of this might have an effect on the results:

Each pixel of the Red Mysterium sensor occupies 29sq. micron. This means the vertical and horizontal length of each pixel is 0.0053851648071345040312507104915403mm. It is from this number that 2048 pixels gives us 11.028817525011464256001455085568mm which was rounded up to 11.03. It is also how I have measured the vertical length which contains 1152 pixels. The 29sq. micron pixel size taken directly from Red.com could however already been a rounded number from 0.0054mm length X 0.0054mm height. In which case each pixel would be 29.16sq Micron. The resulting difference in size for the 2K area would be 11.0592mm length (Compared to 11.03 posted above) and 6.2208 (Compared to 6.20 posted above). The diagonal would be 12.68874529967403829911747653133 or rounded to 12.69mm. Still 0.01mm less than the 12.7mm circle of coverage of 16mm. So by my calculations it would make no difference to the result (I hope they are correct!).

Some of the finest still lenses compare favourably to their Cine lens counterparts and for years companies such as Panavision, ARRI, Samcine, Century Optics and many others have been rehousing still lenses for use in cinema production. Furthermore films as recent as The Bourne Ultimatum have been shot using modified Nikkor lenses (View Article at International Cinematographers Guild). With care and attention to the choice of lens and it’s quality, still lenses can be used productively in a cine environment and get exceptional results.

Common problems do however exist. Firstly, the focus throw on still lenses is smaller than that of cine lenses. When attaching a follow focus to the lens it would require gearing in order to give the focus puller more movement in the focus knob compared to the movement of the focus on the lens. The focus barrel on still lenses more than often rotates in the wrong direction and focus marks are less accurate than on cine lenses.

Another common problem is controlling the aperture. Many still lenses control it via the body of the camera and subsequently have no physical control on the lens itself.

Cine and still lenses have different types of mounts. PL being one of the most commonly adopted cine mounts and Nikon F-Mount and Canon EF-Mount representing two of the most recognised still mounts. Usually these lenses need a complete overhaul and rehousing in order to correctly position the lens to the film plain or sensor of the camera. This is because the flange focal distance in still lenses is smaller than cine lenses and therefore a PL mount conversion cannot simply be added.

Generally speaking still lenses are usually not accurate enough for motion picture work even though they may be optically sound.

In order to combat the inherent problems special mounts for the RedOne, designed and manufactured by Birger Engineering Inc. Boston – Massachusetts, allow for the use of still lenses fitted with either Canon EF, Sigma SA, FourThirds (Olympus/Kodak/Leica & Others) or Nikon F mounts.

By coupling with the lenses automatic controls (Normally used by the corresponding DSLR camera bodies) the mounts can access the motors that control the lenses focus and aperture functions.

A focus puller can control the attached lens using a standard follow focus wheel (Prototype pictured above). By avoiding contact with the focus and aperture controls on the lens itself much of the perils of handling still lenses in a cine environment are removed.

The focus and aperture of a lens can also be controlled remotely, making it ideal for steadicam, handheld or crane work for example.

A further intended advantage of the mount is the lens data information can be converted from it’s native language to Cooke Optics’s /i or Broadcast, which will eventually be included in the camera. This would give the focus puller and post production live accurate data to work from.

Because there is no follow focus on a rig using such a mount, lenses can be swapped out quicker, allowing for greater efficiency on a shoot.

Breathing can be an issue in some still lenses (Although not all Cine lenses can lay claim to being free of it, especially older ones). But given the resolution in any frame shot on the RedOne a relatively simple animated mask can be applied in post to remove any breathing judged to be unsightly. There is also talk of a geared motor being used in conjunction with zoom lenses that will remove any breathing by adjusting the zoom slightly. The motor and the mount would be mapped to the specific lens so that as it is focussed the zoom would also be altered to counter the lenses natural breathing.

The mounts have yet to undergo real world testing so whether still lenses, used in conjunction with the Birger Mounts, are capable enough for use in demanding productions remains to be seen.

There have always been reasons for choosing cine over still lenses and for the large part reliability in the optics we know and trust will, and should, continue. Technology such as Birger’s mounts and the RedOne Digital Cinema Camera could however allow us to select from a wider range of lenses whilst lessening the traditional compromises and expense of rehousing them.

In April 2007 Cooke Optics, Leicester (UK), announced that Red would be adopting their /i Technology in the RedOne Digital Cinema Camera (See press release).

/i means lenses with “intelligence” and Cooke has been implementing their technology in their line of S4 lenses. /i compatible lenses will now send comprehensive, continuous lens data information to the RedOne camera.

The data includes vital information on lens setting, focusing distance, aperture and depth-of-field, hyperfocal distance, serial number, owner data, lens type and focal length in both metric and footage measurements. For zoom lenses, the zoom position is displayed.

Display of the data has previously required Cinematography Electronics’ /i Lens Display System and Cooke’s /i datalink. The information available from /i lenses on the RedOne will help focus pullers by supplying accurate live data via the camera as well as aid post production by including the information within the metadata of the shot footage.

As well as the inclusion within the camera body, Red’s own line of cine lenses include the /i technology. Many other companies have also taken on Cooke’s /i technology. They include Aaton, Avid, Arri, Cinematography Electronics, Dalsa, Preston Cinema Systems, CMotion, Pixel Farm, Service Vision and Mark Roberts Motion Control. The move towards standardization will hopefully help to streamline productions.